(* *)
(***********************************************************************)
-(*
- Time-stamp: <Last modified on 2013-04-22 17:26:34 CEST by Kim Nguyen>
-*)
-
open Format
+let time f arg msg =
+ let t1 = Unix.gettimeofday () in
+ let r = f arg in
+ let t2 = Unix.gettimeofday () in
+ let time = (t2 -. t1) *. 1000. in
+ Logger.msg `STATS "%s: %fms" msg time;
+ r
+
+
+let compose_parallel run auto_list tree nodes () =
+ match auto_list with
+ [ auto ] -> [run auto tree nodes]
+ | _ -> assert false
+
+let compose_sequential run auto_list tree nodes () =
+ [ List.fold_left (fun acc auto ->
+ run auto tree acc) nodes auto_list ]
+
+
+let restart_parallel run auto_list tree nodes () =
+ match auto_list with
+ [ auto ] -> List.map snd (run auto tree nodes)
+ | _ -> assert false
+
+let restart_sequential run auto_list tree nodes () =
+ List.map (fun auto -> run auto tree nodes) auto_list
+
let main () =
let () = Options.parse () in
let doc =
- let fd = open_in !Options.input_file in
- let d = Naive_tree.load_xml_file fd in
- close_in fd; d
+ let fd, close_fd = match !Options.input_file with
+ None | Some "-" | Some "/dev/stdin" -> stdin, ignore
+ | Some input ->
+ let fd = open_in input in fd, fun () -> close_in fd
+ in
+ let d = time Naive_tree.load_xml_file fd "parsing xml document" in
+ close_fd (); d
+ in
+ let queries =
+ time
+ (fun l ->
+ List.map (fun q ->
+ Xpath.Parser.parse
+ (Ulexing.from_utf8_string q)) l)
+ !Options.queries
+ "parsing XPath queries"
in
- let query =
- Xpath.Parser.parse (Ulexing.from_latin1_string !Options.query)
+ (* parallel, compose -> action
+ true, true -> Ata.concat of all automata and single run
+ true, false -> Ata.merge of all automata and single run
+ false, true -> Eval first, then run on results then ...
+ false, false -> Eval first on root, then second on root then ...
+ *)
+ let auto_list =
+ time
+ (fun l ->
+ List.map (fun query -> Xpath.Compile.path query) l)
+ queries
+ "compiling XPath queries"
in
- let auto =
- Xpath.Compile.path query
+ let auto_list =
+ if !Options.parallel then
+ match auto_list with
+ fst :: rest ->
+ let f =
+ if !Options.compose then
+ Ata.concat
+ else
+ Ata.merge
+ in
+ let big_auto = List.fold_left f fst rest in
+ [big_auto]
+ | _ -> assert false
+
+ else
+ auto_list
in
let output =
match !Options.output_file with
| Some f -> open_out f
in
if !Options.stats then begin
- fprintf err_formatter "@[STATS: Query: %a @]@." Xpath.Ast.print_path query;
- fprintf err_formatter "@[STATS: @[Automaton: @\n";
- Ata.print err_formatter auto;
- fprintf err_formatter "@]@]@.";
+ List.iter (fun query ->
+ Logger.msg `STATS "Query: %a " Xpath.Ast.print_path query) queries;
+ List.iter (fun auto ->
+ Logger.msg `STATS "@[Automaton: @\n%a@]" Ata.print auto) auto_list;
end;
- let module Naive = Eval.Make(Naive_tree) in
- let t1 = Unix.gettimeofday() in
- let results = Naive.eval auto doc (Naive_tree.root doc) in
- let teval = (Unix.gettimeofday () -. t1) *. 1000. in
- let t1 = Unix.gettimeofday () in
- output_string output "<xml_result>\n";
- if !Options.count then begin
- output_string output (string_of_int (List.length results));
- output_char output '\n';
- end else
- List.iter (fun n ->
- Naive_tree.print_xml output doc n;
- output_char output '\n'
- ) results;
- output_string output "</xml_result>\n";
- let tprint = (Unix.gettimeofday () -. t1) *. 1000. in
- flush output;
- if output != stdout then close_out output;
- if !Options.stats then begin
- fprintf err_formatter "@[STATS: evaluation time: %fms@]@." teval;
- fprintf err_formatter "@[STATS: serialization time: %fms@]@." tprint
- end
+
+ let module Naive = Run.Make(Naive_tree) in
+ let result_list =
+ let root = [ Naive_tree.root doc] in
+ let f, msg =
+ match !Options.parallel, !Options.compose with
+ true, true ->
+ compose_parallel Naive.eval auto_list doc root, "parallel/compose"
+ | true, false ->
+ restart_parallel Naive.full_eval auto_list doc root, "parallel/restart"
+ | false, true ->
+ compose_sequential Naive.eval auto_list doc root , "sequential/compose"
+ | false, false ->
+ restart_sequential Naive.eval auto_list doc root, "sequential/restart"
+ in
+ time f () ("evaluating query in " ^ msg ^ " mode")
+ in
+ time (fun () ->
+ let count = ref 1 in
+ List.iter (fun results ->
+ output_string output "<xml_result num=\"";
+ output_string output (string_of_int !count);
+ output_string output "\" >\n";
+ if !Options.count then begin
+ output_string output (string_of_int (List.length results));
+ output_char output '\n';
+ end else
+ List.iter (fun n ->
+ Naive_tree.print_xml output doc n;
+ output_char output '\n'
+ ) results;
+ output_string output "</xml_result>\n";
+ incr count
+ ) result_list;
+ flush output;
+ if output != stdout then close_out output
+
+ ) () "serializing results"
let () =
with
Arg.Bad msg -> eprintf "Error: %s\n%!" msg; Options.usage (); exit 1
| Sys_error msg -> eprintf "Error: %s\n%!" msg; exit 2
- | Tree.Parse_error msg -> eprintf "Error: file %s, %s\n%!" !Options.input_file msg; exit 3
+ | Tree.Parse_error msg ->
+ eprintf "Error: %s, %s\n%!"
+ (match !Options.input_file with
+ Some s -> ("file " ^ s)
+ | None -> "[stdin]") msg; exit 3
| Xpath.Ulexer.Error (s, e, msg) -> eprintf "Error: character %i-%i: %s\n%!" s e msg; exit 4
| e -> eprintf "FATAL ERROR: %s\n%!" (Printexc.to_string e); exit 128